llm-router

What This Skill Does

The llm-router skill acts as a high-performance, unified intelligence gateway for your OpenClaw agents. By consolidating 70+ AI models behind a single OpenAI-compatible API, it eliminates the need to manage dozens of individual API keys or rewrite code when switching between model providers like OpenAI, Anthropic, Google, and Deepseek. It simplifies your infrastructure by providing a single endpoint for all inference tasks, complete with built-in fallbacks, intelligent routing, and centralized billing.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/bowen-dotcom/aisa-llm-router-skill After installation, ensure your environment variable is set: export AISA_API_KEY="your-key-here".

Use Cases

• Production-Grade Fallbacks: Configure your agent to query a premium model like GPT-4, and if it fails due to rate limits or downtime, automatically trigger a secondary request to Claude or Gemini to ensure service continuity.
• Cost-Effective Routing: Dynamically route simple chat interactions to high-speed, low-cost models (like gpt-4o-mini or gemini-flash) while escalating complex reasoning tasks to high-end models like Deepseek-R1.
• A/B Testing AI Logic: Compare how different model families respond to the exact same prompt to identify which underlying intelligence performs best for your specific agent tasks.

Example Prompts

1. "Use the llm-router to compare the coding performance of deepseek-v3 and gpt-4o for a simple python script generator."
2. "If the primary LLM fails to process the current prompt, use the fallback strategy to route the request to claude-3-5-sonnet."
3. "Route all user-facing chat messages to gemini-2.0-flash to save costs, but use gpt-4o for any image analysis tasks."

Tips & Limitations

• Model Availability: Always verify supported models via the official marketplace documentation, as providers frequently update their catalogs.
• Temperature Control: While the router handles standard OpenAI parameters, be aware that some smaller or specialized models may have varying sensitivity to temperature settings.
• Latency: While the router is optimized for low latency, remember that choosing larger, reasoning-heavy models will naturally increase total response time. Monitor your usage logs within the AIsa dashboard to track costs and performance trends effectively.